FX vs DX Image Quality Comparison (yes.. this again) Your Thoughts?

donaldejose

You can make some interesting comparisons with this chart:
1. At ISO 100 both the D750 and the D800 have superior dynamic range of 11.38 compared to the D7200s 10.94 or the D810s 11.02.
2. Put only two cameras on the chart, a D4 in DX mode and a D7200. Now shoot both at ISO 100. What is the difference in dynamic range? The D4 in DX is 9.69 while the D7200 is 10.94. If you don't need 10 fps and a large buffer which would you rather shoot at ISO 100 in DX mode? Take the D4 in DX mode off the chart and substitute a D4 in FX mode. At 100 ISO the D4 FX dynamic range is 10.31, still lower than the 10.94 dynamic range of the DX D7200. D4 dynamic range doesn't exceed the D7200 until you reach ISO 200. It sure seems that DX wins the "best bang for the buck" award.

spraynpray

I will add here that the images off the M 4/3rds bodies in our club are well and truly good enough to win prizes and I have yet to hear any of the qualified and experienced judges comment in any way that would lead any body to rush off out and swap to FX.

@WestEndFoto: Don't take it personally, it was never intended to be. I am a Nikon fanboy btw.

Also +1 to DaveyJ's post above.

Paperman

I guess it is obvious for some but can anyone explain why the crop (DX) mode of same (FX) camera gives less DR ?

WestEndFoto

I will add here that the images off the M 4/3rds bodies in our club are well and truly good enough to win prizes and I have yet to hear any of the qualified and experienced judges comment in any way that would lead any body to rush off out and swap to FX.

@WestEndFoto: Don't take it personally, it was never intended to be. I am a Nikon fanboy btw.

Also +1 to DaveyJ's post above.

All is forgiven and no real offense was really taken.

Ironheart

The short answer is smaller sensors gather less light, so therefore a decreased DR. A DX crop is functionally equivalent to a DX camera. I suggest reading all of the material on the site I posted, but this page is the heart of it:
http://home.comcast.net/~NikonD70/GeneralTopics/Sensors_&_Raw/Sensor_Analysis_Primer/DX_Crop_Mode_Photographic_Dynamic_Range.htm

spraynpray

Anybody want to make some stipulations as to the conditions for these comparison images?

Ironheart

Well, I'll start. How about an urban or suburban landscape. Should have a few trees, a few hard edges, contrast btwn dark and light. Hair or eyes are always nice too, perhaps one of our furry friends? I'm not too worried, as long as same subject, with FL and DoF normalized naturally.

heartyfisher

Re test .. I would like to see an object with multi colour and rough detail. Use the Zoom not at the edges of its range to try to equalise the lense IQ as much as possible.

Ironheart

I think we decided 70-200 2.8 @ 100 & 150.
Just for fun, here's Thom's comparison of the D600 vs D7100. As I said however, I think the D7200 is a game changer in terms of DR&IQ
http://www.dslrbodies.com/newsviews/april-2013-nikon-newsviews/dx-versus-fx-again.html

heartyfisher

F2.8 is a weak aperture IQ wise.. I would say F4, F5.6 and F8 would be better

Ironheart

I was only specifying which lens, not what aperture to shoot. While we're on it, what ISO?

WestEndFoto

I was only specifying which lens, not what aperture to shoot. While we're on it, what ISO?

100 or the lowest you can go if you go if you can't go to 100. I would not attach a significance to the test otherwise.

Paperman

@Ironheart

The short answer is smaller sensors gather less light, so therefore a decreased DR. A DX crop is functionally equivalent to a DX camera. I suggest reading all of the material on the site I posted, but this page is the heart of it:

I did give it a try but honestly but what the article explains is way beyond my brain capacity ...( That being said, the graph itself is a beautiful tool. )

I was always under the impression pixel density/size was the reason behind the better DR in FF ... And I thought - not sure if I will be able to express it correctly - everything ( noise/color depth/DR ) happened in the individual photon/pixel etc level and not in millions of them combining to give a result.

Ironheart

@paperman, you need to understand the nature of photons at the quantum level. We can't just count individual ones, and even if we could, there would still be quantum noise in the system. The signal-to-noise ratio determines the DR of the system as a whole. The way I think about it, there is no "dynamic range" of one pixel, i.e it is zero mathematically. Now if you have two pixels, you can have a range, but a very small one. if you have millions, you can have nice curve. One point makes, err well a point. Two points make a line, 3 or more points make a curve.

If we think about an ideal sensor where photon noise is the only issue, the larger the sensor the greater the dynamic range. Dynamic range is the difference between the point at which the sensor becomes saturated and the point at which any detail is lost to noise in the shadows and the ability to represent that in a single image.

A larger sensor will either have larger pixels, or more pixels. Larger pixels mean a greater capacity to store photons (all else being equal) and more light being captured per pixel, which gives more light in the shadows, and therefore a greater dynamic range. More pixels means similar noise per pixel but more pixels to average over to reduce shadow noise, and therefore increased dynamic range.

You can plot the "ideal" curves on the graphs to see how close or far a sensor is from the ideal, which is ultimately limited by the sensor size.

http://www.cambridgeincolour.com/tutorials/dynamic-range.htm
http://www.imatest.com/docs/dynamic/

spraynpray

There is no "dynamic range" of one pixel, i.e it is zero mathematically. Now if you have two pixels, you can have a range, but a very small one. if you have millions, you can have nice curve. One point makes, err well a point. Two points make a line, 3 or more points make a curve.

If we think about an ideal sensor where photon noise is the only issue, the larger the sensor the greater the dynamic range. Dynamic range is the difference between the point at which the sensor becomes saturated and the point at which any detail is lost to noise in the shadows and the ability to represent that in a single image.

Am I misunderstanding what you are saying Ironheart?

Surely the difference between the point where the output from a pixel drops below its noise level and the point where it is saturated is a function of each individual pixel and that inconsistency in performance between each pixel is shown by (is what gives) the 'grain' or noise in our low light images (non-linearity)? It may become less noticeable viewed small when it is a high megapixel sensor, but it is in fact worse when viewed large than a lower megapixel sensor i.e. the difference between a D810 and a D750. I can't see how the number of pixels in a sensor influence each pixel's ability to continue to output current for photon strikes in a predictable manner with increasing light levels. What I am saying is that I cannot see how the performance of each individual pixel is dependant on, or related to, the number or area of pixels around it. I can see that the DR can improve with an increase in pixel size and that improvements in signal to noise ratio and sensitivity can be achieved by new tech or manufacturing process improvements, but not simply by increasing the number of pixels.

heartyfisher

Seriously we need a new thread just for that graph ..

Ironheart

It's a generalization. Paperman was asking for a non-mathematical explanation. If you pixel-peep at a RAW file, you are still looking at a combination of adjacent pixels, due to the demosaic algorithm. The "inconsistency" between pixels is not a performance issue, but a combination of quantum noise, electrical noise, and analog to digital conversion noise, and will never be zero. Noise reduction algorithms, which are run even on RAW files by the camera, take the neighboring pixels into consideration. Once you convert to JPEG all pixels are tied together.

Heartyfisher, how is this not related to the topic?

Paperman

It is hard for me to understand how a ( FF ) sensor when cropped ( to DX mode ) loses IQ ( even though, I guess the guy has exactly proved that in his tests ) . It is like taking an image, cropping it on PC and then expecting it to lose DR/Color depth/noise performance ...( yeah, I know , it is a bit too much of an over-simplificaton :-) )

Ironheart

@paperman, I was going to comment exactly on this, so thanks for the segue :-) What I was specifically addressing in answering @sevencrossing is comparing a DX crop of the same FX sensor. It should be obvious that a crop of any picture will increase noise. Try it for yourself. By increasing noise, you are automatically decreasing DR, as the two are inexorably linked.

Perhaps we do need a separate thread...

Paperman

If I am looking at 1:1 magnification of either the original or the cropped image, shouldn't I be seeing the exact same thing ? I have simply cut from sides of the picture .

sevencrossing

Perhaps we do need a separate thread...

+!
How about

PIXEL PEEPING

I quite like this definition

Attempting to ascertain the image quality yielded by a digital camera or lens by close inspection of its output at 100% zoom (or greater). Some consider this behaviour to be obsessive.

Ironheart

By definition PDR is not pixel peeping. If you read this page from Bill you should understand. This is my last post on this subject on this thread :-B
http://home.comcast.net/~NikonD70/GeneralTopics/Sensors_&_Raw/Sensor_Analysis_Primer/Photographic_Dynamic_Range_Summary.htm

DaveyJ

Sounds like what Ironheart is using (800mm equivalent is what I need next! What Nikon I is that on. Probably not my Nikon I AW I but I don't get to use that much anymore as it is the job site camera nowadays. Perhaps the NEXT camera investment for ME is what Ironheart is putting in an overhead carry on... And I am going to bet that I would MAYBE able to buy only a few pieces of that gear.....Oh RETIREMENT where you work more hours at 1/20th the income!!!

Msmoto

FX vs DX Image Quality Comparison.....

Large format vs. small format......

The discussion is interesting in that we all seem to have our bias, and every opinion is valid. My conclusion is that to compare one with another is an impossibility, like apples and oranges. Apple pie...vs. orange? pie. There are clear advantages to having a compact camera, yet some of us prefer a larger format, for a lot of reasons....

In the old days of motorsports photography, a 210mm lens on a 4"x5" was common. Today, one cannot position themselves ten feet from a car moving 100 mph, and we also have the capability to optically bring ourselves back into this position....on a smaller format camera.

So, how does one discuss all this? I do not know. My very compact Olympus 4/3rds format camera produces some stunning images. But, I prefer what I get from FX, because that is what I have.

It is my thinking, there is no "quality difference" between FX and DX as the primary difference is simply the size of the format. The glass is what will determine the IQ, in spite of all the pixel peeping discussion. But, once again, it is the individual who is behind the camera which determines the final result. And, this is about 90% of the final result. Thus, format differences may account for 10% of the difference, but probably not much more.

And, of course, this is one opinion only....mine....based upon simply being old....LOL

Ironheart

@DaveyJ Nikon 1V3, 10, 18.5, 32mm primes, 6.7-13, 10-100, 70-300 zooms, is what I have with me. As to how this relates to the topic, I would say crop sensors, whether 2.7 or 1.5 factor have the size/weight advantage. I'll post some images later to show that the IQ trade-off is manageable and worth it. To me.

I'm not even sure how I would get the equiv FX gear here, unless I was willing to check it as luggage. Not something I'd want to do, unless I was a pro and had someone else to worry about logistics, proper insurance, and a Sherpa, I mean assistant.